delta/lua-patterns
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

thread-safe error handling (in theory); panic if pattern is bad

Browse source
This commit is contained in:
Steve Donovan 2017-04-18 15:44:02 +02:00
parent c8dd33f165
commit 9be8354722
2 changed files with 56 additions and 50 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

59
src/lib.rs
View file

@ -49,8 +49,7 @@ impl <'a> LuaPattern<'a> {
err_msg, self.matches.as_mut_ptr()) as usize; err_msg, self.matches.as_mut_ptr()) as usize;
let ep = *err_msg; let ep = *err_msg;
if ! ep.is_null() { if ! ep.is_null() {
let slice = CStr::from_ptr(ep); panic!(format!("lua-pattern {:?}",CStr::from_ptr(ep)));
println!("{:?}", slice);
} }
} }
@ -77,16 +76,16 @@ impl <'a> LuaPattern<'a> {
self.matches(text); self.matches(text);
vec.clear(); vec.clear();
for i in 0..self.n_match { for i in 0..self.n_match {
vec.push(&text[self.bounds(i)]); vec.push(&text[self.capture(i)]);
} }
self.n_match > 0 self.n_match > 0
} }
pub fn range(&self) -> ops::Range<usize> { pub fn range(&self) -> ops::Range<usize> {
self.bounds(0) self.capture(0)
} }
pub fn bounds(&self, i: usize) -> ops::Range<usize> { pub fn capture(&self, i: usize) -> ops::Range<usize> {
ops::Range{ ops::Range{
start: self.matches[i].start as usize, start: self.matches[i].start as usize,
end: self.matches[i].end as usize end: self.matches[i].end as usize
@ -96,7 +95,7 @@ impl <'a> LuaPattern<'a> {
pub fn first_match<'b>(&mut self, text: &'b str) -> Option<&'b str> { pub fn first_match<'b>(&mut self, text: &'b str) -> Option<&'b str> {
self.matches(text); self.matches(text);
if self.n_match > 0 { if self.n_match > 0 {
Some(&text[self.bounds(if self.n_match > 1 {1} else {0})]) Some(&text[self.capture(if self.n_match > 1 {1} else {0})])
} else { } else {
None None
} }
@ -112,7 +111,7 @@ impl <'a> LuaPattern<'a> {
let mut res = String::new(); let mut res = String::new();
while self.matches(slice) { while self.matches(slice) {
// full range of match // full range of match
let all = self.bounds(0); let all = self.range();
let captures = Captures{m: self, text: slice}; let captures = Captures{m: self, text: slice};
let repl = lookup(captures); let repl = lookup(captures);
// append everything up to match // append everything up to match
@ -133,7 +132,7 @@ pub struct Captures<'a,'b> {
impl <'a,'b> Captures<'a,'b> { impl <'a,'b> Captures<'a,'b> {
pub fn get(&self, i: usize) -> &'b str { pub fn get(&self, i: usize) -> &'b str {
&self.text[self.m.bounds(i)] &self.text[self.m.capture(i)]
} }
pub fn num_matches(&self) -> usize { pub fn num_matches(&self) -> usize {
@ -184,7 +183,7 @@ impl <'a,'b>Iterator for GMatch<'a,'b> {
None None
} else { } else {
let first = if self.m.n_match > 1 {1} else {0}; let first = if self.m.n_match > 1 {1} else {0};
let slice = &self.text[self.m.bounds(first)]; let slice = &self.text[self.m.capture(first)];
self.text = &self.text[self.m.range().end..]; self.text = &self.text[self.m.range().end..];
Some(slice) Some(slice)
} }
@ -200,32 +199,32 @@ mod tests {
#[test] #[test]
fn captures_and_matching() { fn captures_and_matching() {
let mut m = LuaPattern::new("(one).+"); let mut m = LuaPattern::new("(one).+");
assert_eq!(m.captures(" one two"),&["one two","one"]); assert_eq!(m.captures(" one two"), &["one two","one"]);
let empty: &[&str] = &[]; let empty: &[&str] = &[];
assert_eq!(m.captures("four"),empty); assert_eq!(m.captures("four"), empty);
assert_eq!(m.matches("one dog"),true); assert_eq!(m.matches("one dog"), true);
assert_eq!(m.matches("dog one "),true); assert_eq!(m.matches("dog one "), true);
assert_eq!(m.matches("dog one"),false); assert_eq!(m.matches("dog one"), false);
let text = "one dog"; let text = "one dog";
let mut m = LuaPattern::new("^(%a+)"); let mut m = LuaPattern::new("^(%a+)");
assert_eq!(m.matches(text),true); assert_eq!(m.matches(text), true);
assert_eq!(&text[m.bounds(1)], "one"); assert_eq!(&text[m.capture(1)], "one");
assert_eq!(m.matches(" one dog"),false); assert_eq!(m.matches(" one dog"), false);
// captures without allocation // captures without allocation
let captures = m.match_captures(text); let captures = m.match_captures(text);
assert_eq!(captures.get(0),"one"); assert_eq!(captures.get(0), "one");
assert_eq!(captures.get(1),"one"); assert_eq!(captures.get(1), "one");
let mut m = LuaPattern::new("(%S+)%s*=%s*(.+)"); let mut m = LuaPattern::new("(%S+)%s*=%s*(.+)");
// captures as Vec // captures as Vec
let cc = m.captures(" hello= bonzo dog"); let cc = m.captures(" hello= bonzo dog");
assert_eq!(cc[0], "hello= bonzo dog"); assert_eq!(cc[0], "hello= bonzo dog");
assert_eq!(cc[1],"hello"); assert_eq!(cc[1], "hello");
assert_eq!(cc[2],"bonzo dog"); assert_eq!(cc[2], "bonzo dog");
// captures as iterator // captures as iterator
let mut iter = m.match_captures(" frodo = baggins").into_iter(); let mut iter = m.match_captures(" frodo = baggins").into_iter();
@ -234,8 +233,6 @@ mod tests {
assert_eq!(iter.next(), Some("baggins")); assert_eq!(iter.next(), Some("baggins"));
assert_eq!(iter.next(), None); assert_eq!(iter.next(), None);
} }
#[test] #[test]
@ -258,12 +255,24 @@ mod tests {
#[test] #[test]
fn gsub() { fn gsub() {
use std::collections::HashMap;
let mut m = LuaPattern::new("%$(%S+)"); let mut m = LuaPattern::new("%$(%S+)");
let res = m.gsub("hello $dolly you're so $fine!", let res = m.gsub("hello $dolly you're so $fine!",
|cc| cc.get(1).to_uppercase() |cc| cc.get(1).to_uppercase()
); );
assert_eq!(res,"hello DOLLY you're so FINE!"); assert_eq!(res,"hello DOLLY you're so FINE!");
let mut map = HashMap::new();
map.insert("dolly", "baby");
map.insert("fine", "cool");
map.insert("good-looking", "pretty");
let mut m = LuaPattern::new("%$%((.-)%)");
let res = m.gsub("hello $(dolly) you're so $(fine) and $(good-looking)",
|cc| map.get(cc.get(1)).unwrap_or(&"?").to_string()
);
assert_eq!(res,"hello baby you're so cool and pretty");
} }
} }

47
src/lua-str.c
View file

@ -14,11 +14,6 @@ typedef struct LuaMatch {
} LuaMatch; } LuaMatch;
// not thread safe (yet)
static jmp_buf s_jmp_buf;
static char s_msg_buff[256];
/* macro to `unsign' a character */ /* macro to `unsign' a character */
#define uchar(c) ((unsigned char)(c)) #define uchar(c) ((unsigned char)(c))
@ -47,6 +42,8 @@ typedef struct MatchState {
const char *init; const char *init;
ptrdiff_t len; ptrdiff_t len;
} capture[LUA_MAXCAPTURES]; } capture[LUA_MAXCAPTURES];
jmp_buf jump_buf;
char msg_buff[256];
} MatchState; } MatchState;
/* recursive function */ /* recursive function */
@ -55,21 +52,19 @@ static const char *match (MatchState *ms, const char *s, const char *p);
#define L_ESC '%' #define L_ESC '%'
#define SPECIALS "^$*+?.([%-" #define SPECIALS "^$*+?.([%-"
// error handling, hm?? NB static int throw_error(MatchState *ms, const char *fmt,...) {
static int throw_error(const char *fmt,...) {
va_list ap; va_list ap;
va_start(ap,fmt); va_start(ap,fmt);
vsnprintf(s_msg_buff,sizeof(s_msg_buff),fmt,ap); vsnprintf(ms->msg_buff,sizeof(ms->msg_buff),fmt,ap);
va_end(ap); va_end(ap);
longjmp(s_jmp_buf,1); longjmp(ms->jump_buf,1);
return 0; return 0;
} }
static int check_capture (MatchState *ms, int l) { static int check_capture (MatchState *ms, int l) {
l -= '1'; l -= '1';
if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED) if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
return throw_error("invalid capture index %%%d", l + 1); return throw_error(ms,"invalid capture index %%%d", l + 1);
return l; return l;
} }
@ -77,7 +72,7 @@ static int capture_to_close (MatchState *ms) {
int level = ms->level; int level = ms->level;
for (level--; level>=0; level--) for (level--; level>=0; level--)
if (ms->capture[level].len == CAP_UNFINISHED) return level; if (ms->capture[level].len == CAP_UNFINISHED) return level;
return throw_error("invalid pattern capture"); return throw_error(ms,"invalid pattern capture");
} }
@ -86,14 +81,14 @@ static const char *classend (MatchState *ms, const char *p) {
switch (*p++) { switch (*p++) {
case L_ESC: { case L_ESC: {
if (p == ms->p_end) if (p == ms->p_end)
throw_error("malformed pattern (ends with '%')"); throw_error(ms,"malformed pattern (ends with '%')");
return p+1; return p+1;
} }
case '[': { case '[': {
if (*p == '^') p++; if (*p == '^') p++;
do { /* look for a `]' */ do { /* look for a `]' */
if (p == ms->p_end) if (p == ms->p_end)
throw_error("malformed pattern (missing ']')"); throw_error(ms,"malformed pattern (missing ']')");
if (*(p++) == L_ESC && p < ms->p_end) if (*(p++) == L_ESC && p < ms->p_end)
p++; /* skip escapes (e.g. `%]') */ p++; /* skip escapes (e.g. `%]') */
} while (*p != ']'); } while (*p != ']');
@ -168,7 +163,7 @@ static int singlematch (MatchState *ms, const char *s, const char *p,
static const char *matchbalance (MatchState *ms, const char *s, static const char *matchbalance (MatchState *ms, const char *s,
const char *p) { const char *p) {
if (p >= ms->p_end - 1) if (p >= ms->p_end - 1)
throw_error("malformed pattern " throw_error(ms,"malformed pattern "
"(missing arguments to '%b')"); "(missing arguments to '%b')");
if (*s != *p) return NULL; if (*s != *p) return NULL;
else { else {
@ -218,7 +213,7 @@ static const char *start_capture (MatchState *ms, const char *s,
const char *p, int what) { const char *p, int what) {
const char *res; const char *res;
int level = ms->level; int level = ms->level;
if (level >= LUA_MAXCAPTURES) throw_error("too many captures"); if (level >= LUA_MAXCAPTURES) throw_error(ms,"too many captures");
ms->capture[level].init = s; ms->capture[level].init = s;
ms->capture[level].len = what; ms->capture[level].len = what;
ms->level = level+1; ms->level = level+1;
@ -251,7 +246,7 @@ static const char *match_capture (MatchState *ms, const char *s, int l) {
static const char *match (MatchState *ms, const char *s, const char *p) { static const char *match (MatchState *ms, const char *s, const char *p) {
if (ms->matchdepth-- == 0) if (ms->matchdepth-- == 0)
throw_error("pattern too complex"); throw_error(ms,"pattern too complex");
init: /* using goto's to optimize tail recursion */ init: /* using goto's to optimize tail recursion */
if (p != ms->p_end) { /* end of pattern? */ if (p != ms->p_end) { /* end of pattern? */
switch (*p) { switch (*p) {
@ -285,7 +280,7 @@ static const char *match (MatchState *ms, const char *s, const char *p) {
const char *ep; char previous; const char *ep; char previous;
p += 2; p += 2;
if (*p != '[') if (*p != '[')
throw_error("missing '[' after '%f' in pattern"); throw_error(ms,"missing '[' after '%f' in pattern");
ep = classend(ms, p); /* points to what is next */ ep = classend(ms, p); /* points to what is next */
previous = (s == ms->src_init) ? '\0' : *(s - 1); previous = (s == ms->src_init) ? '\0' : *(s - 1);
if (!matchbracketclass(uchar(previous), p, ep - 1) && if (!matchbracketclass(uchar(previous), p, ep - 1) &&
@ -357,14 +352,13 @@ static void push_onecapture (MatchState *ms, int i, const char *s,
if (i >= ms->level) { if (i >= ms->level) {
if (i == 0) { /* ms->level == 0, too */ if (i == 0) { /* ms->level == 0, too */
mm->start = 0; mm->start = 0;
mm->end = e - s ; mm->end = e - s;
//lua_pushlstring(ms->L, s, e - s); /* add whole match */
} else } else
throw_error("invalid capture index"); throw_error(ms,"invalid capture index");
} }
else { else {
ptrdiff_t l = ms->capture[i].len; ptrdiff_t l = ms->capture[i].len;
if (l == CAP_UNFINISHED) throw_error("unfinished capture"); if (l == CAP_UNFINISHED) throw_error(ms,"unfinished capture");
if (l == CAP_POSITION) { if (l == CAP_POSITION) {
mm[i].start = ms->capture[i].init - ms->src_init + 1; mm[i].start = ms->capture[i].init - ms->src_init + 1;
mm[i].end = mm[i].start; mm[i].end = mm[i].start;
@ -384,7 +378,6 @@ static int push_captures (MatchState *ms, const char *s, const char *e, LuaMatch
return nlevels; /* number of strings pushed */ return nlevels; /* number of strings pushed */
} }
int str_match (const char *s, unsigned int ls, const char *p, unsigned int lp, char **err_msg, LuaMatch *mm) { int str_match (const char *s, unsigned int ls, const char *p, unsigned int lp, char **err_msg, LuaMatch *mm) {
const char *s1 = s; const char *s1 = s;
MatchState ms; MatchState ms;
@ -392,9 +385,13 @@ int str_match (const char *s, unsigned int ls, const char *p, unsigned int lp, c
if (anchor) { if (anchor) {
p++; lp--; /* skip anchor character */ p++; lp--; /* skip anchor character */
} }
memset(ms.msg_buff,0,sizeof(ms.msg_buff));
if (setjmp(s_jmp_buf) != 0) { if (setjmp(ms.jump_buf) != 0) {
if (err_msg != NULL) *err_msg = s_msg_buff; if (err_msg != NULL) {
*err_msg = strdup(ms.msg_buff);
}
return 0; return 0;
} }